On 04.10.2017 17:48, Laszlo Ersek wrote:
Hi,
On 09/21/17 16:28, Auger Eric wrote:
Hi David,
On 20/09/2017 16:34, David Hildenbrand wrote:
On 20.09.2017 16:31, Gerd Hoffmann wrote:
Hi,
Dropping from os section:
<loader readonly="yes"
type="pflash">/usr/share/edk2.git/ovmf-x64/OVMF_CODE-pure-
efi.fd</loader>
<nvram
template="/usr/share/edk2.git/ovmf-x64/OVMF_VARS-pure-
efi.fd">/var/lib/libvirt/qemu/nvram/fedora-org-drm-qxl-
base_VARS.fd</nvram>
Bad idea I guess.
1) Does the assert trigger right at startup? Or how is it triggered?
Yes, right at startup, before OVMF is done initializing.
So probably something in OVMF triggers it.
I'll try to include ovmf to reproduce it.
2) Does your setup work when dopping the assertion?
Can try tomorrow.
I encounter the problem on ARM too. My setup works when dropping the
assertion.
Thanks
Eric
I think that assertion might not be stable, because properties (romd
mode) might not be stable and can change. So if it works without the
assert, dropping it is the right thing to do.
There seems to be a new issue report on LaunchPad about the same
(f357f564be0b) commit, stating that the removal of the assertion does
not help:
https://bugs.launchpad.net/qemu/+bug/1721221
(Just trying to connect the dots here; I have no comments on the commit
otherwise. CC'ing Joe who has filed the LP report.)
Thanks
Laszlo
Hi Joe,
do you have some time to test with the following branch?
https://github.com/davidhildenbrand/qemu/tree/kvm_slot
If it works - very good. If not, can you give me what QEMU prints (I
added some debug output).
Looks something like this:
set_phys_mem(0): start: 0x0, size: 0x8000000, add: 1
Add(0): start: 0x0, size: 0x8000000, ram: 0x0x7f6697e00000
set_phys_mem(0): start: 0xfffc0000, size: 0x40000, add: 1
Add(0): start: 0xfffc0000, size: 0x40000, ram: 0x0x7f66a7200000
set_phys_mem(0): start: 0x0, size: 0x8000000, add: 0
...